sign in sign up
<<
Home , Agaricomycetidae, Ammoniphilus, Asanoa, Bionectriaceae, Chaetothyriales, Chaetothyriomycetidae

The ISME Journal (2015) 9, 2671–2681 © 2015 International Society for Microbial Ecology All rights reserved 1751-7362/15 www.nature.com/ismej ORIGINAL ARTICLE Soil-foraging animals alter the composition and co-occurrence of microbial communities in a desert shrubland

David J Eldridge1, Jason N Woodhouse2, Nathalie JA Curlevski2,3, Matthew Hayward4,5, Mark V Brown2 and Brett A Neilan2 1Centre for Ecosystem Science, School of Biological, Earth and Environmental Sciences, University of NSW, Sydney, NSW, Australia; 2School of Biotechnology and Biomolecular Sciences, University of NSW, Sydney, NSW, Australia; 3Faculty of Science, Aquatic Processes Group, University of Technology, Ultimo, NSW, Australia; 4Australian Wildlife Conservancy, Nichols Point, VIC, Australia and 5School of Environment, Natural Resources and Geography; and School of Biological Science, Bangor University, Bangor, UK

Animals that modify their physical environment by foraging in the soil can have dramatic effects on ecosystem functions and processes. We compared bacterial and fungal communities in the foraging pits created by bilbies and burrowing bettongs with undisturbed surface soils dominated by biocrusts. Bacterial communities were characterized by and , and fungal communities by and Archaeosporomycetes. The composition of bacterial or fungal communities was not observed to vary between loamy or sandy soils. There were no differences in richness of either bacterial or fungal operational taxonomic units (OTUs) in the soil of young or old foraging pits, or undisturbed soils. Although the bacterial assemblage did not vary among the three microsites, the composition of fungi in undisturbed soils was significantly different from that in old or young foraging pits. Network analysis indicated that a greater number of correlations between bacterial OTUs occurred in undisturbed soils and old pits, whereas a greater number of correlations between fungal OTUs occurred in undisturbed soils. Our study suggests that digging by soil-disturbing animals is likely to create successional shifts in soil microbial and fungal communities, leading to functional shifts associated with the decomposition of organic matter and the fixation of nitrogen. Given the primacy of organic matter decomposition in arid and semi-arid environments, the loss of native soil-foraging animals is likely to impair the ability of these systems to maintain key ecosystem processes such as the mineralization of nitrogen and the breakdown of organic matter, and to recover from disturbance. The ISME Journal (2015) 9, 2671–2681; doi:10.1038/ismej.2015.70; published online 1 May 2015

Introduction with exotic pests including the European rabbit (Oryctolagus cuniculus) and predation by introduced Australia has suffered one of the highest rates of such as the domestic cat (Felis catus) and the global mammal extinctions over the past 200 years red fox (Vulpes vulpes) (Johnson, 2006). Two species since European settlement (Woinarski et al., 2012). that have suffered substantial range restrictions are Losses have been most pronounced in the critical – the greater bilby (Macrotis lagotis) and the burrowing weight range (35 5500 g) mammals, which were bettong (Bettongia lesueur). Recent attempts have once common over large areas of continental been made to reintroduce these animals into Australia (Johnson, 2006). The loss of these animals, predator-proof exclosures within their former range or the contraction of their ranges, has been attributed in an effort to re-establish viable populations (James to multiple causes associated with European and Eldridge, 2007). settlement and pastoral practices such as altered Many of Australia’s locally extinct animals forage fire regimes, overgrazing by livestock, competition extensively in the soil for seeds, bulbs, invertebrates and fungi (Robley et al., 2001; James et al., 2011; Eldridge et al., 2012). Foraging disturbs the soil Correspondence: DJ Eldridge, Centre for Ecosystem Science, School surface and breaks up the surface crust (biocrust), of Biological, Earth and Environmental Sciences, University of altering rates of water infiltration, and creating small NSW, Sydney, NSW 2052, Australia. E-mail: [email protected] pits and depressions that trap water, soil, organic Received 5 October 2014; revised 5 February 2015; accepted matter and seed (James et al., 2009). These pits 25 March 2015; published online 1 May 2015 develop into patches of higher nutrients, with greater Animal foraging alters microbial community DJ Eldridge et al 2672 concentrations of plant-available nitrogen and This compares with young pits (o3 months old), carbon than the surrounding soil matrix (James which have relatively low levels of litter and organic et al., 2010) and often a different vegetation commu- matter (D J Eldridge, unpublished data). We would nity (Lavelle et al., 2006). Studies worldwide have expect pit age to influence microbial community shown that modification of the abiotic environment structure, as these old pits (~12 months) would have by these animals, a process referred to as ecosystem more time to establish seedlings and accumulate engineering (sensu Jones et al., 1994), alters litter and microorganisms that are present on energy flows and resource availability, increases adjacent, undisturbed surfaces. Furthermore, older species richness, diversity and productivity, through pits could provide a greater range of different niche construction, ultimately controlling the environments, with differences in depth, shape availability and distribution of resources to other and orientation, and therefore different soil chem- organisms (for example, Whitford and Kay, 1999; istry and organic matter at varying stages of Jones et al., 2010). decomposition. An important process moderated by soil- We compared the community structure of soil disturbing animals in arid environments is the microbial communities in old and young pits with decomposition of organic matter. Litter and organic the undisturbed surface soil on two soil types in an matter in these systems are spatially and temporally area where bilbies and bettongs have been reintro- variable, and is often concentrated within the duced into their former range. Both bilbies and foraging pits of animals (James and Eldridge, 2007). bettongs construct pits while foraging for buried Litter is a source of carbon, nitrogen and other trace seed, invertebrates and plant roots. The pits of these elements, and provides habitat for a range of micro- two species are indistinguishable, and range from and macro-invertebrates involved in the decomposi- cylindrical-shaped excavations about 15-cm wide tion of organic matter (Haslem et al., 2011). Litter and up to 20-cm deep to shallow basin-like struc- falling into pits comes into close contact with soil, tures (Eldridge et al., 2012). Pits are constructed only where it is held in situ more effectively than if it once, and unlike cache pits of heteromyid rodents remained on the soil surface where it is subject to (Geluso, 2005), are rarely reworked. Because pits removal by wind and water (Whitford, 2002). vary in depth and shape, and are constructed in soils Together with reduced evaporation resulting from of different texture, they provide a range of different lower temperatures in the pits than the undisturbed physical environments that influences the trapping surface (Eldridge and Mensinga, 2007), this increases and retention of litter and the breakdown of organic the time period over which soil moisture is optimum matter. for microbial decomposition and nutrient minerali- We hypothesized that the microbial community in zation (Steinberger and Whitford, 1983; Jacobsen and pit soils would support more microorganisms com- Jacobsen, 1998; Whitford, 2002). Litter remaining on monly associated with decomposing litter. Conver- the surface, however, is subject to photodegradation sely, we expected that the microbial community (Austin and Vivanco, 2006), potentially reducing the composition in undisturbed soils would support a return of carbon to the soil organic pool. community dominated by cyanobacteria, given the Soil-disturbing animals therefore play an impor- extensive cover of biocrusts on the soil surface. We tant role in bringing surface-resident organic matter used microbial network analysis to examine the into contact with soil microorganisms. The accumu- structure of microbial communities, particularly in lation of litter in the pits is also likely to exert a relation to resilience and reactivity (Ruiz-Moreno strong selective pressure on microorganisms essen- et al., 2006; Bissett et al., 2013). Examination of tial for the decomposition process. Given the marked microbial networks improves our understanding of differences in the biotic (litter cover and composi- why undisturbed soils might be resistant to nutrient tion) and abiotic (for example, surface temperature, amendment, how microbial community structure is soil moisture) environments between pits and altered following pit construction, and how digging undisturbed soils, that is, those soils undisturbed promotes nutrient enrichment within these micro- by animal activity (for example, Vossbrinck et al., sites (James et al., 2009). 1979; Wallwork et al., 1985; Eldridge and Mensinga, 2007), we expected that the pits would differ in the composition of soil microorganisms. For example, Materials and methods studies of foraging disturbances constructed by the short-beaked echidna (Tachyglossus aculeatus) indi- The study area cate a greater diversity and abundance of micro- Our study was undertaken within the Australian arthropods and higher rates of microbial respiration Wildlife Conservancy’s Scotia Sanctuary in south- in the pits than undisturbed soil (Eldridge and western, New South Wales, Australia (33°43’S, Mensinga, 2007), suggesting that there are differ- 143°02’E) where locally extinct bilbies and bettongs ences in the abundance or structure of microbial have been released into predator-proof exclosures. communities. Over time, pits collect organic matter, Soil samples were collected from two systems; and research has shown that pits over about (i) mallee (Eucalyptus spp.) west-east-trending 12 months old have high levels of organic carbon. dunes of Quaternary alluvium characterized by

The ISME Journal Animal foraging alters microbial community DJ Eldridge et al 2673 calcareous and siliceous sands (Rudosols) and and fungal specific tag-encoded FLX amplicon (ii) the inter-dunal swales and plains extending to pyrosequencing of each sample was carried out these dunes, which are up to 500-m wide, comprised using the primers 27 f/519r and funSSUF/funSSUR mainly of loamy, calcareous soils (Calcarosols). The respectively (Lucero, 2011) on a Roche GS-FLX vegetation on the dunes is moderately dense mallee Titanium at the Research and Testing Laboratory (Eucalyptus socialis, E. dumosa) and the plains (Lubbock, TX, USA). Sequence reads were analyzed are dominated by open mallee woodland with using MOTHUR v1.22 (www.mothur.org) software scattered belah (Casuarina pauper) and sugarwood package (Schloss et al., 2009). Initial quality proces- (Myoporum platycarpum), and a variable cover of sing of 454 sequence reads was performed using the shrubs such as punty bush (Senna artemisioides), mothur implementation of PyroNoise (Quince et al., hopbush (Dodonaea viscosa), turpentine (Eremo- 2011) using default settings. Sequences containing phila sturtii), pinbush wattle (Acacia burkittii) and o200 bp, containing ambiguous bases and homo- assorted bluebushes (Maireana spp.), depending on polymers longer than 8 bp in length were removed. whether trees had been removed. Shrubs covered The remaining sequences were aligned to either the about 50% of the area of the plains. The climate in the bacterial or fungal alignments of the SILVA release area is semi-arid, with cool winters (mean ⩽ 17 °C) and 102 reference alignment. Chimeric sequences were hot summers (mean 30 °C). Rainfall is highly spa- identified and removed using the mothur implemen- tially and temporally variable and averages 243 mm tation of uchime (Edgar et al., 2011). The taxonomic per year. Rainfall is evenly distributed between the identity of each unique sequence was determined by six warmer months and the six cooler months. comparison against the SILVA release 102 reference database. Taxonomic assignment was made at each level, given a bootstrap value greater than 80, using Field sampling the RDP classifier (Wang et al., 2007). Sequences that The location, size, depth and age of all foraging pits failed to be classified at the level or constructed by bilbies and bettongs have been were classified as either Mitochondria, Archaea or monitored at 36 large sites at the Scotia Sanctuary Eukaryota/Prokaryota in the respective datasets, since 2007. Because sites were visited every were removed. Sub-sampling was performed at a 3 months, we were able to calculate the relative age level of 400 sequences per sample for the bacterial of particular pits. In October 2009, we collected soil dataset and 1300 sequences per sample for the fungal samples from six sites: three on sandy dunes and dataset. Implementation of this process resulted in three on loamy plains. At each of the six sites, we the exclusion of a bacterial young loam soil sample sampled three microsites: (i) young foraging pits, that and bacterial young sand soil sample, as these is, pits constructed since the previous measurements samples contained fewer than the 400 sequences (o3 months old), (ii) old foraging pits, that is, pits required. To ensure a balanced design across the older than 12 months and (iii) undisturbed non-pit bacterial dataset, the corresponding samples were surface soils at least 3 m from any pit. At each of the subsequently excluded from the bacterial old pit soil six sites, we sampled each microsite at 10 locations. and bacterial surface pit soil sets (2 replicate sites of For the young pits, soil was removed from the 2 soil types × 3 microsites. Uncorrected pairwise uppermost 10 mm layer of the soil surface or from distances were calculated between sequence reads the base of the pits after removing any existing with the final clustering of operational taxonomic organic material. Biocrust was not removed from the units (OTUs) performed at a 0.03 distance threshold soil prior to sampling. Approximately 5 g of soil was using the average neighbor algorithm (Schloss and collected with a sterilized spatula. The material from Westcott, 2011). The identity of each OTU defined at the 10 locations was then bulked and stored on ice 0.03 distance threshold was obtained from the before being transported back to the laboratory. The consensus of each sequence within that OTU at a same procedure was used to collect samples from old confidence threshold of 80. From these data, two pits and undisturbed surfaces. This resulted in a individual data matrices were generated, one for database of 18 bulked samples (3 replicate sites of 2 and one for fungi, each matrix containing soil types × 3 microsites). every OTU and the number of reads assigned to it from each sample. In this instance, the relative proportion of each OTU was used as a proxy for Molecular analysis abundance, as absolute abundance measures were Environmental DNA was isolated from 500 mg of soil not obtained. using the FASTDNA Spin Kit for Soil (MP Bio Laboratories, Inc., Carlsbad, CA, USA) according to the manufacturer’s instructions and stored at − 80 °C until use. DNA was quantified using a NanoDrop Statistical analysis ND-1000 Spectrophotometer (Thermo Fisher Scientific, We used permutation multivariate analysis of Waltham, MA, USA) and the quality checked by variance (PERMANOVA; Anderson et al., 2008) to PCR amplification of the 16 S rRNA gene using the examine differences in the composition of a data primer pair 27 f/519r (Weisburg, 1991). Bacterial matrix of 2500 bacterial OTUs, defined at 0.03

The ISME Journal Animal foraging alters microbial community DJ Eldridge et al 2674 distance threshold, and a data matrix of 5895 fungal Results OTUs, defined at 0.03 distance threshold, in relation to microsite (undisturbed soils, young pits, old pits) Richness of bacterial and fungal taxa and soil type (loam, sand). Relative abundance data Most bacterial and fungal OTUs occurred at very low were used after resampling, to ensure an equivalent abundances, with a substantial number of abun- number of sequences. The first stratum of this dances equal to one. Of the original 2500 bacterial analysis considered soil type and the second stratum OTUs after resampling, 320 (14%) contributed 50% microsite and its interaction with soil type. Pairwise of total OTU abundances. For fungal OTUs, 525 (9%) a posteriori comparisons were made, where neces- of the 5895 OTUs contributed 50% of total fungal sary, using a multivariate analogue of the t statistic, abundances. There were no differences in bacterial OTU richness (that is, different number of OTUs) the probability levels being obtained by permutation. – We tested for differences in richness and diversity of among the different soils (P = 0.24; range 238 332 taxa with a mixed-model ANOVA with the same OTUs) or among the three microsites (P=0.47). Similarly, fungal richness did not vary with soil structure as the PERMANOVA analysis. Richness – and diversity data were checked for homogeneity of texture (P = 0.81; range 397 873) or among the three variance (Levene’s test) and normality using diag- microsites (P = 0.17). nostic tests but no transformations were needed. For all analyses, significant differences between means were examined using Fisher’s Protected Least Community composition of bacterial and fungal taxa Significant Difference test. The procedure was Bacterial communities were observed to contain repeated for the fungal data. a high proportion of Actinobacteria, and to a The degree of association of OTUs with respect to lesser extent, Alphaproteobacteria and Acidobacteria microsite was measured with Indicator Species (Figure 1a). Cyanobacteria appeared to constitute a Analysis in R (De Cáceres et al., 2012) using a data large proportion of the bacterial community, parti- matrix consisting of 2500 bacterial OTUs and 5895 cularly in undisturbed soils. Fungal communities fungal OTUs. Indicator values combine information were observed to contain a high proportion of on relative abundance and frequency of species, and Lecanoromycetes, and to a lesser extent, Archae- the indicator value is maximal (IV = 100%) when all osporomycetes (Figure 1b). individuals of a given species are restricted to a There was no significant difference in the compo- particular microsite (for example, old pit), and all sition of either bacterial or fungal OTUs assemblages samples from the particular microsite contain an between loamy and clay soils (P40.30). The compo- occurrence of that species. Data (at the OTU level) sition of the bacterial assemblage did not vary among were randomized among the treatments and a Monte the three microsites (P=0.21; Figure 2a), but there Carlo randomization procedure performed with 1000 was a significant effect for fungi (Pseudo F2,8 = 3.08, iterations to determine the statistical significance of P(Perm) = 0.003). The composition of fungi in undis- the indicator values. turbed soils was significantly different from that in The degree of association of OTUs with respect to old (pairwise t = 2.14, P = 0.029) or young (t = 2.02, one another within each microsite was measured P = 0.02) pits, but there was no significant differences using the Pearson’s correlation coefficient (r). between old and young pits (P=0.47; Figure 2b). Bacterial and fungal OTU tables, defined at 0.03 distance threshold, were separated on the basis of microsite, and then reduced by removing any OTUs Microsite indicators that did not occur across at least 75% of available Six cyanobacterial OTUs (Gp I (3 OTUs), Gp X, Gp samples. A Pearson’s r score and P-value were VII and an unclassified OTU) were indicators of calculated pairwise for each bacterial OTU using undisturbed pits, as were the single Asanoa the rcor.test algorithm, available from the ltm OTU (Actinobacteria), a Segetibacter OTU (Sphingo- package (available from https://cran.r-project.org/ bacteria) and an unclassified alphaproteobacterial package=ltm) as implemented in R version 3.0.2. OTU. A single Hylangium OTU (Myxobacteria, Delta- For each correlation, P-values were generated and ), OTU (Rhizobiales, Alpha- the false discovery rate was maintained below 5% proteobacteria) and a Gp IV actinobacterial OTU using the Benjamini-Hochberg procedure (Benjamini were indicators of old pits. A single Rubrobacter and Hochberg, 1995). Visualization of these interac- OTU (Actinobacteria), OTU (, tions, incorporating taxonomic, abundance and ) and Actinaurispora OTU (Actino- microsite occurrence information, was made with bacteria) were indicators of young pits (Table 1). the freely available Cytoscape package version 2.8.3 Overall, fungal taxa were better discriminators of the (available at: www.cytoscape.org). For each network, three microsites, with 20 orders containing 170 topological metrics of connectivity and density were OTUs, with indicator values 40.70, and almost calculated using the network analysis plug-in exclusively from sub-phylum . These (Assenov et al., 2008). Networks pre-embedded with included orders (genera Columnosphaeria, sample and OTU-specific information are provided Delphinella), ( Glyphium), Leca- in the Supplementary Material. norales (genera Sphaerophoraceae, Cladoniaceae),

The ISME Journal Animal foraging alters microbial community DJ Eldridge et al 2675

Figure 2 Multi-dimensional scaling biplot of the first two Figure 1 Relative abundance of major (a) bacterial and (b) fungal dimensions of an ordination of a reduced matrix of (a) 280 taxa within each microsite. Larger circles indicate greater bacterial OTUs and (b) 135 fungal OTUs. Note the clustering of abundance. undisturbed samples for both bacteria and fungi.

Myxotrichaceae (genus ), pits = 0.088) and centralization (undisturbed = 0.089, ()andPleosporales (genera , old pits = 0.096) (Table 3). , ). Ten fungal genera Within the fungal networks, the highest mean (particularly Eupenicillium, Hamigera, number of correlations between OTUs (20.497) was and an unclassified taxon from the family Bulgaria) observed in undisturbed soils, where many more were highly indicative (IV40.81) of young pits. Old OTUs (1814 OTUs) were present across multiple pits contained a wide range of different OTUs, with the samples than in young (321) or old (485) pit soils. orders Chaetothyriales, Dothideales, , Similar to the bacterial networks, young pit , Mycocaliciales and having soils returned the lowest values for density a large number of OTUs that were strongly indicative (0.009), but were also the most centralized (0.116). (IV469%)ofolderpits(Table2). Old pit soils were the least clustered (0.472) and the least centralized (0.067). Undisturbed soils were similar to young pit soils in terms of clustering Network analysis (undisturbed = 0.652, young pits = 0.647), whereas Within the bacterial networks, the mean number of fungal young and old pit soils were only similar in correlations between OTUs was greater in old pit relation to the mean number of correlations soils (3.45) than either undisturbed (2.516) or young between OTUs. pit (1.294) soils, consistent with a larger number of OTUs co-occurring across the samples (Table 3). The majority of the associations present in young pit soils Discussion were between a small number of alphaproteobacter- ial and actinobacterial OTUs. Young pit soils Soil foraging by semi-fossorial animals in arid areas returned the lowest values for network metrics of disrupts surface crusts, alters rates of water infiltra- clustering (0), density (0.081) and centralization tion, and creates small pits and depressions that trap (0.050). Undisturbed soils and old pit soils were water, soil, organic matter and seed (James and similar in relation to clustering (undisturbed = 0.566, Eldridge, 2007). We expected to detect substantial old pits = 0.547), density (undisturbed = 0.084, old differences in the soil microbial community between

The ISME Journal Animal foraging alters microbial community DJ Eldridge et al 2676 Table 1 Bacterial taxa, to the level of genus, that are significantly associated with different microsites using Indicator Species Analysis

Order Family Genus Microsite IV P-value No of OTUs

Cyanobacteria Family I Group I Undisturbed 0.866 0.047 3 Cyanobacteria Family X Group X Undisturbed 0.866 0.046 1 Cyanobacteria Unclassified Unclassified Undisturbed 0.866 0.049 1 Actinomycetales Asanoa Undisturbed 0.866 0.046 1 Cyanobacteria Family VIII Group VIII Undisturbed 0.866 0.047 1 Sphingobacteriales Chitinophagaceae Segetibacter Undisturbed 0.866 0.049 1 Alphaproteobacteria Alphaproteobacteria Alphaproteobacteria Undisturbed 0.866 0.047 1 Myxococcales Cystobacteraceae Hyalangium Old 1.000 0.008 1 Acidobacteria Acidobacteria Group IV Old 0.913 0.030 1 Rhizobiales Microvirga Old 0.812 0.028 1 Rubrobacterales Rubrobacteraceae Rubrobacter Young 0.905 0.012 1 Ammoniphilus Young 0.866 0.040 1 Actinomycetales Micromonosporaceae Actinaurispora Young 0.866 0.040 1

Abbreviations: IV, indicator value; OTUs, operational taxonomic units.

intact undisturbed soils and recently excavated or Compositional differences between pit and undisturbed older, more established pits in response to differ- soils ences in plant and litter cover, organic matter Consistent with information from arid soils world- decomposition and soil nutrient concentrations. wide, the bacterial community contained high Although we detected some significant differences proportions of Actinobacteria and Alphaproteo- in the fungal community composition between the bacteria (Figure 1a) (Yeager et al., 2004; Kuske, 2012). soil surface and the pits (described below), there At the community level, we detected no significant were no discernible differences in the bacterial differences in bacterial community composition community and in the fungal community between between pits and undisturbed soils (Figure 2). young and old pits, largely because of the high However, consistent with our first hypothesis, variability among microsites (Figure 1). Conse- filamentous diazotrophic (Cyanobacteria GpI), baeo- quently, we undertook an analysis that would test cystous (Cyanobacteria GpVIII) and unicellular whether the physical variability that is observed in (Cyanobacteria GpX) cyanobacteria were found to pit soils, in regard to moisture and nutrient trapping, be indicators of undisturbed soils (Table 1) with a influenced the occurrence of individual species or reduction in the observed abundance of cyanobac- the manner in which individual species exhibited terial sequence reads when soils were disturbed correlations with one another. Indicator species (Figure 1, Table 1). Cyanobacteria were present in analysis was implemented to identify specific OTUs undisturbed soils, however, the presence of these that were more strongly associated with a particular taxa as indicators was reflective of both a decrease in microsite type. Critically, indicator species analysis the abundance of cyanobacterial groups and a shift has been previously shown to be suitable for within the morphological and physiological nature identifying variable taxa where there was no of cyanobacteria between undisturbed and pit soils. prior assessment, or no significant variation, in Among the heterotrophic population, actinobacterial the larger community composition (De Cáceres and members of the Rubrobacteridae that are pioneers in Legendre, 2009; De Cáceres and Legendre, 2009; biological crust formation (Yeager et al., 2004) De Cáceres et al., 2012). That it were possible to dominated both undisturbed and pit soils, with a identify species that were statistically indicative of single Rubrobacter OTU, an indicator of young pit particular microsites when the multivariate soils. In addition to cyanobacterial groups, the (PERMANOVA) analysis was insignificant high- Sphingobacterial genus Segetibacter has been lights the fact that there is substantial hetero- previously affiliated with the decomposition of geneity within microsites, and suggests a level cyanobacteria- and plant-derived phytodetritus of functional redundancy within microbial taxa (Li et al., 2011). that prevents large-scale perturbation of the Fungal communities in undisturbed and pit soils community despite the loss of species. On the comprised a wide range of saprotrophs, with basis of the indicator species (De Cáceres and Lecanoromycetes, the largest class of lichenized Legendre, 2009; De Cáceres et al., 2012) and fungi, and to a lesser extent, Archaeosporomycetes, microbial network (Chaffron, 2010; Bissett et al., comprising about 80% of sequences across the three 2013) analyses, there is sufficient evidence to microsites (Figure 1b). Along with Pezizomycotina, suggest that pits may be associated with a reduction these fungal taxa perform a diverse array of ecologi- in autotrophic groups (Figure 1, Tables 1 and 2) cal functions including wood and litter decomposi- that are compensated for by an emergence of tion, mycorrhizal associations and symbioses, taxa capable of decomposing organic material animal and plant pathogens (Spatafora et al., 2006). (Tables 1 and 2) and reduced resilience in the Evidence for active recession, or at least competitive microbial communities (Table 3). inhibition, of microbial groups from the old pits was

The ISME Journal Animal foraging alters microbial community DJ Eldridge et al 2677 Table 2 Fungal taxa, to the level of genus, that are significantly associated with different microsites using Indicator Species Analysis

Subclass Order Family Genus Microsite IV P No of OTUs

Eurotiomycetidae Trichocomaceae Unclassified Undisturbed 0.94 0.002 1 Pleosporales Leptosphaeria Undisturbed 0.91 0.006 8 Pleosporomycetidae Pleosporales Trematosphaeria Undisturbed 0.91 0.006 4 Skeletonemataceae Skeletonema Unclassified Unclassified Undisturbed 0.91 0.002 1 Strombidiidae Strombidium Unclassified Unclassified Undisturbed 0.91 0.007 1 Dothideales Delphinella Undisturbed 0.90 0.003 3 Naviculaceae Navicula Unclassified Unclassified Undisturbed 0.90 0.003 4 Pleosporomycetidae Pleosporales Phaeosphaeria Undisturbed 0.88 0.013 6 Pleosporomycetidae Pleosporales Undisturbed 0.88 0.011 4 Dothideomycetidae Dothideales Dothioraceae Columnosphaeria Undisturbed 0.86 0.012 17 Chaetothyriales Chaetothyriales Glyphium Undisturbed 0.86 0.012 19 Pleosporomycetidae Pleosporales Undisturbed 0.86 0.011 2 Chaetothyriomycetidae Chaetothyriales Chaetothyriales Sarcinomyces Undisturbed 0.86 0.011 2 Mycocaliciomycetidae Mycocaliciales Sphinctrina Undisturbed 0.86 0.012 11 Letendraea Unclassified Undisturbed 0.86 0.012 10 Leotiomycetes Myxotrichaceae Myxotrichaceae Geomyces Undisturbed 0.85 0.012 6 Sporadotrichida Halteriidae Halteria Unclassified Undisturbed 0.85 0.014 7 Dothideomycetes Botryosphaeria Undisturbed 0.82 0.016 1 Lecanorales Lecanorineae Cladoniaceae Undisturbed 0.82 0.016 1 Dothideomycetidae Dothideales Dothideales Hortaea Undisturbed 0.82 0.015 2 Xylariomycetidae Xylariales Undisturbed 0.82 0.023 1 Xylariomycetidae Xylariales Undisturbed 0.82 0.012 1 Helotiales Bulgariaceae Bulgaria Unclassified Undisturbed 0.82 0.012 1 Dothideomycetes Kirschsteiniothelia Unclassified Unclassified Undisturbed 0.82 0.016 2 Pezizales Pezizaceae Peziza Unclassified Undisturbed 0.82 0.016 1 Sordariomycetidae Magnaporthales Magnaporthaceae Pseudohalonectria Undisturbed 0.81 0.021 3 Lecanoromycetidae Lecanorales Lecanorineae Sphaerophoraceae Undisturbed 0.81 0.018 5 Agaricomycetidae Lycoperdaceae Lycoperdon Undisturbed 0.76 0.039 1 Dothideomycetidae Dothideales Dothioraceae Delphinella Old 0.91 0.004 1 Eurotiales Trichocomaceae Chromocleista Old 0.82 0.015 1 Chaetothyriomycetidae Chaetothyriales Chaetothyriales Glyphium Old 0.82 0.025 6 Dothideomycetes Tubeufiaceae Letendraea Unclassified Old 0.82 0.025 1 Chaetothyriomycetidae Chaetothyriales Chaetothyriales Sarcinomyces Old 0.80 0.025 1 Mycocaliciomycetidae Mycocaliciales Sphinctrinaceae Sphinctrina Old 0.80 0.027 3 Sporadotrichida Halteriidae Halteria Unclassified Old 0.80 0.024 2 Dothideomycetidae Dothideales Dothioraceae Columnosphaeria Old 0.79 0.025 5 Pleosporomycetidae Pleosporales Leptosphaeriaceae Leptosphaeria Old 0.79 0.030 4 Hypocreales Old 0.75 0.034 2 Xylariomycetidae Xylariales Amphisphaeriaceae Pestalosphaeria Old 0.74 0.045 1 Eurotiomycetidae Eurotiales Trichocomaceae Hamigera Young 0.87 0.011 4 Hypocreomycetidae Hypocreales Bionectriaceae Bionectriaceae Young 0.86 0.013 2 Dothideomycetidae Dothideales Dothioraceae Columnosphaeria Young 0.86 0.013 1 Eurotiomycetidae Eurotiales Trichocomaceae Eupenicillium Young 0.84 0.016 5 Leotiomycetes Myxotrichaceae Myxotrichaceae Geomyces Young 0.82 0.021 1 Mycocaliciomycetidae Mycocaliciales Sphinctrinaceae Sphinctrina Young 0.82 0.012 1 Pleosporomycetidae Pleosporales Leptosphaeriaceae Leptosphaeria Young 0.79 0.041 2 Dothideomycetes Tubeufiaceae Letendraea Unclassified Young 0.79 0.026 2

Abbreviations: IV, indicator value; OTUs, operational taxonomic units. Only taxa with an IV40.75 are shown.

Table 3 Metrics obtained from analysis of scale-free microbial networks of bacterial and fungal microbial communities

Taxon and microsite OTUs Edges Mean number of neighbours Clustering coefficient Density Centralization

Bacteria Old 40 70 3.450 0.547 0.088 0.096 Undisturbed 31 39 2.516 0.566 0.084 0.089 Young 17 11 1.294 0 0.081 0.050

Fungi Old 135 485 7.185 0.472 0.054 0.067 Undisturbed 177 1814 20.497 0.652 0.166 0.095 Young 81 321 7.926 0.647 0.009 0.116

Abbreviation: OTUs, operational taxonomic units. Edges represent the number of significant positive and negative Pearson correlation coefficients identified following implementation of the Benjamini-Hochberg procedure at a minimum false discovery rate of 5%.

The ISME Journal Animal foraging alters microbial community DJ Eldridge et al 2678 found, with the insect and plant pathogenic fungi, Myrangium (Smith, 1948) and the epiphytic mela- Delphinella, Leptosphaeria, Trematosphaeria and nized taxon Sarcinomyces (Wollenzien et al., 1997). Columnosphaeria, found almost exclusively in The Lecanorales are predominantly lichen-forming undisturbed and young pit soils. Glomeromycetes, fungi that are mycobionts of the genera Xanthopar- which comprise arbuscular mycorrhizal species, melia, Parmotrema and Xanthoria, which are com- represented about 3% of sequences in young pits mon corticolous of Callitris glaucophylla and 2% of sequences in old pit and trees that occur in the study area (Filson and Rogers, undisturbed soils. 1979). These taxa are typically found on the soil surface or in the pits on detached plant material. At some sites, we also recorded the vagant lichen Community development with pit age Chondropsis semiviridis from within the pits. This Rubrobacter, Ammoniphilus and Actinaurispora lichen, which has no attachment to the soil, moves were the only bacterial indicators of young pits and freely along the surface by wind action (Eldridge likely represent remnants of the sub-surface com- and Leys, 1999). Similarly, Cladonia spp., another munity. Rubrobacter is a cosmopolitan and abundant common soil lichen genus, was found on undis- taxon in arid zone soils (Yeager et al., 2004). The turbed surfaces. Along with the lichen genera presence of Ammoniphilus and Actinaurispora in and , it is one of the most young pit soils is likely due to the deposition of plant common lichens forming biocrusts on stable soils in material. Amminophilus has been reported as a arid and semi-arid areas (Eldridge and Koen, 1998). strictly aerobic oxalotroph utilizing plant- and Despite our inability to discriminate between the algae-derived oxalic acid as a sole carbon. Actinaur- bacterial community of old and young pits, we ispora are known plant endophytes, inhabiting recorded three indicator species, Hyalangium and Camptotheca acuminata species (Zhu et al., 2012). Microvirga, and a Gp IV Acidobacteria. The two The family Micromonosporacaeae, to which Acti- proteobacterial species were indicative of the pre- naurispora belongs, however, has been tentatively sence of established vascular plants. Hyalangium, correlated with increasing moisture content in arid belongs to the group of Myxobacteria that uses plant and semi-arid soils (Bachar et al., 2010), which may lignin and produces small bioactive molecules. contribute to the presence of this species as an Microvirga has been implicated in nodule formation, indicator of young soils. Trichocomaceae species facilitating nitrogen-fixing processes within the were the primary fungal indicators of young pits. A rhizosphere (Ardley et al., 2012). The occurrence of single Trichocomaceae species was a key fungal these groups in old pit soils is likely to enhance indicator of undisturbed soils, suggesting that fungal nitrogen fixation, presumably to levels greater than communities of young pits contain residual surface those in the undisturbed and young pit soils, and taxa prior to the colonization and diversification of support the growth of vascular plants occurring in fungal communities observed in older pits. Tricho- these microsites. comaceae species are predominantly saprotrophic, have aggressive colonization strategies and a high tolerance to extreme environmental conditions such Microbial co-occurrence in pit and undisturbed soils as soil drying, high temperature and metal toxicity Our analyses thus far indicate that initial disturbance (Houbraken and Samson, 2011). Their presence in reduces the abundance of key photoautotrophic young pits could indicate opportunistic colonization groups, and that over extended periods of time, of recently disturbed soil. capture of organic matter leads to changes in the On the basis of the criteria used to select the abundance of some taxa, with increases in those taxa microsites, progression of the microbial community likely reflecting an increased capacity for the from young to old pits occurs over a period of assimilation of organic carbon and nitrogen matter. 9–12 months. Over this time, although little change Resilience is the ability of a system to recover from occurred within the microbial community composi- large disturbances, typically over short time frames. tion between pit stages, a discernable difference was Reactivity, however, is the capacity of a system to observed between the undisturbed and pit bacterial respond to small perturbations over extended peri- and fungal communities, irrespective of their age. ods. Under such circumstances, the apparent equili- Microbial richness among microsites, however, brium may appear stable, despite moving to a new remained unchanged. Spore propagule density and steady state over long time periods (Neubert et al., arbuscular mycorrhizal fungi diversity are known to 2009). Modularity, defined by the number and size of decline with increasing tillage associated with groups of highly interconnected nodes within a agriculture (Brito et al., 2010; Schalamuk et al., network, is positively correlated with reactivity, 2013). However, this was not reflected in our fungal and negatively correlated with resilience (Ruiz-Moreno species richness, which remained unchanged over et al., 2006). Analysis of both bacterial and fungal time. The progressive accumulation of fungal species microbial networks revealed stark differences in attached to organic matter and seed in the pits is modularity, reflected in the values of clustering, consistent with the presence of several lichenized density and centralization, of microbial co-occurrence lecanoralean genera including and networks between undisturbed soils and pit soils at

The ISME Journal Animal foraging alters microbial community DJ Eldridge et al 2679 different developmental stages (Table 3). Clustering et al., 2013). It has been speculated that these nodes coefficients and density (network connectivity) represent critical control points in the cycling of scores tending towards a value of 1 indicate a highly nutrients within the system (Ruiz-Moreno et al., modular system whereas those tending towards zero 2006; Bissett et al., 2013). Thus, it is realistic to represent the opposite (Bissett et al., 2013). Low suggest that these centralized taxa act to stabilize the values of clustering and density associated with microbial community. It should be highlighted that microbial communities from contaminated and these observations were made in the context of a reference estuarine sediments indicate historical small number of samples defining each microsite, as community 'stress' contributing to functional redun- well as few sequence reads being available to dancy and reduced correlations among species identify species–species correlations. Our observa- (Sun et al., 2013). This was reinforced by marginally tions between the bacterial and fungal datasets lower values, for each of these metrics, in contami- suggest that these metrics are susceptible to nated sediments, with the suggestion that this sequence depth, and pre-treatment of the data by anthropogenic perturbation has contributed an addi- retaining only semi-ubiquitous (occurring across at tional stress. least 75% of samples) OTUs. These values may also In the present study, bacterial species–species be influenced by the level of heterogeneity within correlations within the young pit soils were almost microsites. Despite this, our analyses of network nonexistent. A clustering coefficient of zero and a metrics from the bacterial communities suggested slightly lower density value were consistent with that the community structure of old pit soils reflect reduced modularity, and an increase in functional that of undisturbed soils. Over the long term, this redundancy associated with a recent external stress would tend towards decreased responses to nutrient (Sun et al., 2013). In contrast, undisturbed and old inputs into these soils. This, however, may be pit soils were more consistent with increased partially offset by frequent deposition of plant matter modularity, suggesting a lack of functional redun- due to the establishment and growth of vascular dancy, with greater species–species correlations, and plants within old pit soils, and subsequent assimila- increased clustering and density. This suggests to us tion of this matter by saprotrophic fungi. that the bacterial community present in undisturbed soils and old pit soils are more reactive and less resilient than young pit soils. Within the fungal Conclusions communities, the number of correlations among species, clustering coefficients and density, and Our study suggests that digging by soil-disturbing hence modularity, were highest in surface soils and animals is likely to create successional shifts in soil lowest in the old pit soils, suggesting that fungal microbial and fungal communities, which could communities within old pits are less reactive and account for increases in organic matter of nitrogen more resilient. In contrast, the young pit soils exhibit in old foraging pits (James et al. 2009). The observed reduced modularity, and increased resilience, sug- richness of fungal and bacterial OTUs, in undis- gesting that they are likely to respond to nutrient turbed soils, and young and old pits did not differ, amendments over the short-term, thereby driving though fewer correlations, and hence an increased large and dramatic structural changes. This is largely resilience, were observed between bacterial OTUs in because of the high degree of physical disturbance young pits, and fungal OTUs in young and old pits. created when foraging pits are established. Within This suggests that these communities are more likely the old pit soils, the bacterial community has largely to respond over the short term to nutrient amend- regained the modularity observed within the undis- ment, thus promoting nutrient enrichment and turbed soils. The fungal community, however, is contributing to a form of patchiness that is critical apparently more resistant at this stage than in the for the functioning of arid systems. The action of soil- undisturbed soils, suggesting it is able to continue to disturbing animals therefore leads to the development drive structural changes in response to events such of a mosaic of different patches with a varying as litter deposition. complement of microorganisms. Given the wide variety A high level of centralization, as a consequence of in pit size, depth, substrate and spatial configuration, the high frequency of centralized nodes, was this differential microbial activity will likely lead to the observed among the fungal community in young pit creation of a mosaic of patches of differing resource soils (Bulgariaceae, Myxotrichaeae, Trichocomaceae, availability, analogous to larger surface-resident biotic Tubeufiaceae) and among the bacterial community in patches such as hummocks and debris mounds. Our undisturbed (Rubrobacteraceae, Geodematophilia- work suggests that microbial community composition ceae, Bradyrhizobiaceae)andoldpit(Rhodobacter- and co-occurrence change with physical disturbance aceae, Bradyrhizobiaceae, Geodermatophiliaceae, during the formation of foraging pits. Given the Beijerinckiaceae, Comamonadaceae, Methylobacteria- primacy of organic matter decomposition in arid and ceae) soil (Supplementary Information). Centralized semi-arid environments, the loss of native soil-foraging nodes have been proposed to represent keystone animals from these systems may well impair the ability species, exhibiting a large influence of the ‘informa- of these systems to maintain key ecosystem processes tion’ transfer throughout the community (Bisssett and to recover from disturbance.

The ISME Journal Animal foraging alters microbial community DJ Eldridge et al 2680 Conflict of Interest Eldridge DJ, Mensinga A. (2007). Foraging pits of the Short-Beaked Echidna (Tachyglossus aculeatus) as The authors declare no conflict of interest. small-scale patches in a semi-arid Australian box woodland. Soil Biol Biochem 39: 1055–1065. Eldridge DJ, Leys JF. (1999). Wind dispersal of the vagant Acknowledgements lichen Chondropsis semiviridisin semi-arid eastern – We thank Ivan Wong for field assistance and Samantha Australia. Aust J Bot 47: 157 164. Travers for comments on an earlier draft of the manuscript. Eldridge DJ, Koen TB. (1998). Cover and floristics of We acknowledge the considerable financial, logistic and microphytic soil crusts in relation to indices of – technical support provided by the Australian Wildlife landscape health. Plant Ecol 137: 101 114. Conservancy through their Scotia Sanctuary. This research Eldridge DJ, Koen TB, Huang N, Killgore A, Whitford WG. was supported by the Australian Government under ARC (2012). Animal foraging as a mechanism for sediment LP0882630. movement and soil nutrient development: evidence from the semi-arid Australian woodlands and the Chihuahuan Desert. Geomorph 157/158: 131–141. Filson RB, Rogers RW. (1979). Lichens of South Australia. Govemment Printer: South Australia. Geluso K. (2005). Benefits of small-sized caches for scatter- References hoarding rodents: influence of cache size, depth, and soil moisture. J Mammal 86: 1186–1192. Anderson MJ, Gorley RN, Clarke KR. (2008). PERMA- Haslem A, Kelly LT, Nimmo DG, Watson SJ, Kenny SA, NOVA+ for PRIMER: guide to software and statistical Taylor RS et al. (2011). Habitat or fuel? Implications of methods. PRIMER-E: Plymouth, UK. long-term, post-fire dynamics for the development of Ardley JK, Parker MA, De Meyer SE, Trengove RD, key resources for fauna and fire. J Appl Ecol 48: O'Hara GW, Reeve WG et al. (2012). Microvirga lupini 247–256. sp. nov., Microvirga lotononidis sp. nov. and Micro- Houbraken J, Samson RA. (2011). Phylogeny of virga zambiensis sp. nov. are alphaproteobacterial and the segregation of Trichocomaceae into three root-nodule bacteria that specifically nodulate and fix families. Studies Mycol 70:1–51. nitrogen with geographically and taxonomically sepa- Jacobsen KM, Jacobsen PJ. (1998). Rainfall regulates rate legume hosts. Int J System Evol Microb 62: decomposition of buried cellulose in the Namib Desert. 2579–2588. J Arid Environ 38: 571–583. Assenov Y, Ramirez F, Schelhorn SE, Lengauer T, James AI, Eldridge DJ. (2007). Reintroduction of fossorial Albrecht M. (2008). Computing topological parameters native mammals and potential impacts on ecosystem of biological networks. Bioinformatics 24: 282–284. processes in an Australian desert landscape. Biol Austin AT, Vivanco L. (2006). Plant litter decomposition in Conserv 138: 351–359. a semi-arid ecosystem controlled by photodegradation. James AI, Eldridge DJ, Hill BM. (2009). Foraging animals Nature 442: 555–558. create fertile patches in an Australian desert shrub- Bachar A, Al-Ashhab A, Soares MIM, Sklarz MY, Angel R, land. Ecography 32: 723–732. Ungar ED, Gillor O. (2010). Soil microbial abundance James AI, Eldridge DJ, Moseby KE. (2010). Foraging pits, and diversity along a low precipitation gradient. litter, and plant germination in an arid shrubland. Microb Ecol 60: 453–461. J Arid Environ 74: 516–520. Benjamini Y, Hochberg Y. (1995). Controlling the false James AI, Eldridge DJ, Koen TB, Moseby KE. (2011). discovery rate: a practical and powerful approach to Can the invasive European rabbit (Oryctolagus multiple testing. J Royal Stat Soc Ser B 57: 289–300. cuniculus) assume the soil engineering role of – Bissett A, Brown MV, Siciliano SD, Thrall PH. (2013). locally-extinct natives? Biol Invas 13: 3027 3038. ’ Microbial community responses to anthropogenically Johnson C. (2006). Australia s mammal extinctions: induced environmental change: towards a systems a 50000 year history. Cambridge University Press: approach. Ecol Lett 16: 128–139. Melbourne. Brito I, Goss MJ, de Carvalho M, Chatagnier O, Jones CG, Lawton JH, Shachak M. (1994). Organisms as ecosystem engineers. Oikos 69: 373–386. van Tuinen D. (2010). Impact of tillage system on Jones CG, Gutierrez JL, Byers JE, Crooks JA, Lambrinos JG, arbuscular mycorrhiza fungal communities in the soil Talley TS. (2010). A framework for understanding under Mediterranean conditions. Soil Till Res 121: – physical ecosystem engineering by organisms. Oikos 63 67. 119: 1862–1869. Chaffron S, Rehrauer H, Pernthaler J, von Mering C. (2010). Kuske CR, Yeager CM, Johnson S, Ticknor LO, Belnap J. A global network of coexisting microbes from environ- (2012). Response and resilience of soil biocrust mental and whole-genome sequence data. Genome Res bacterial communities to chronic physical disturbance – 20: 947 959. in arid shrublands. ISME J 6: 886–897. De Cáceres M, Legendre P. (2009). Associations between Lavelle P, Daecans T, Aubert M, Barot S, Blouin M, species and groups of sites: indices and statistical Bureau F et al. (2006). Soil invertebrates and ecosystem inference. Ecology 90: 3566–3574. services. Eur J Soil Biol 42:S3–S15. De Cáceres M, Legendre P, Wiser SK, Brotons L. (2012). Li H, Xing P, Chen M, Bian Y, Wu QL. (2011). Short-term Using species combinations in indicator analyses. bacterial community composition dynamics in Method Ecol Evol 3: 973–982. response to accumulation and breakdown of Edgar RC, Haas BJ, Clemente JC, Quince C, Knight R. Microcystis blooms. Water Res 45: 1702–1710. (2011). UCHIME improves sensitivity and speed of Lucero ME, Unc A, Cooke P, Dowd S, Sun S. (2011). chimera detection. Bioinformatics 27: 2194–2200. Endophyte microbiome diversity in micropropagated

The ISME Journal Animal foraging alters microbial community DJ Eldridge et al 2681 Atriplex canescens and Atriplex torreyi var griffithsii. anthropogenic contamination over multiple environ- PLoS One 6: e17693. mental gradients. Environ Micro 15: 2517–2531. Neubert MG, Caswell H, Solow AR. (2009). Detecting Vossbrinck CR, Coleman DC, Woolley TA. (1979). Abiotic reactivity. Ecology 90: 2683–2688. and biotic factors in litter decomposition in a semiarid Quince C, Lanzen A, Davenport RJ, Turnbaugh PJ. (2011). grassland. Ecology 60: 265–271. Removing noise from pyrosequenced amplicons. BMC Wallwork JA, Kamill BM, Whitford WG. (1985). Distribu- Bioinf 12: 38. tion and diversity patterns of soil mites and other Robley AJ, Short J, Bradley S. (2001). Dietary overlap microarthropods in a Chihuahuan desert site. J Arid between the burrowing bettong and the European Environ 9: 215–231. rabbit in semi-arid coastal Western Australia. Wildl Wang Q, Garrity GM, Tiedje JM, Cole JR. (2007). Naive – Res 28: 341 349. Bayesian classifier for rapid assignment of rRNA Ruiz-Moreno D, Pascual M, Riolo R. (2006). Exploring sequences into the new bacterial . Appl network space with genetic algorithms: modularity, Environ Micro 73: 5261–5267. resilience and reactivity. In: Pascual M, Dunne JA (eds), Weisburg WG, Barns SM, Pelletier Da, Lane DJ. (1991). Ecological networks: linking structure to dynamics in – 16 S ribosomal DNA amplification for phylogenetic food webs. Oxford University Press: UK, pp 187 208. study. J Bacteriol 173: 697–703. Schalamuk S, Velázquez S, Cabello M. (2013). Dynamics of Whitford WG. (2002). Ecology of Desert Systems. Elsevier arbuscular mycorrhizal fungi spore populations and Science: London. their variability under contrasting tillage systems in Whitford WG, Kay FR. (1999). Biopedturbation by desert at different phonological stages. Biol Agric – – mammals: a review. J Arid Environ 41: 203 230. Hortic 29:38 45. Woinarski JCZ, Burbidge AA, Harrison PL. (2012). The Schloss PD, Westcott SL, Ryanbin T, Hall JR, Hartmann M, Action Plan for Australian Mammals, 2012 CSIRO Hollister EB et al. (2009). Introducing mothur: open- Publishing: Melbourne, Australia. source, platform-independent, community-supported Wollenzien U, Hoog GS, Krumbein WE, Uijthf JMJ. (1997). software for describing and comparing microbial Sarcinomycespetricola, a new microcolonial communities. Appl Environ Micro 75: 7537–7541. from marble in the Mediteranean Basin. Antone van Schloss PD, Westcott SL. (2011). Assessing and improving – methods used in operational taxonomic unit-based Leeuwenhoek 71: 281 288. approaches for 16 S rRNA gene sequence analysis. Yeager CM, Kornosky JL, Housman DC, Grote EE, Appl Environ Micro 77: 3219–3226. Belnap J, Kuske CR. (2004). Diazotrophic community Smith HD. (1948). Biological control of citrus insects. structure and function in two successional stages In: Webber HJ, Batchelor LD (eds), The citrus industry of biological soil crusts from the Colorado Plateau The production of the crop Vol. II. California Press: and Chihuahuan Desert. Appl Environ Micro 70: – USA, pp 597–626. 973 983. Spatafora JW, Sung GH, Johnson D, Hesse C, O'Rourke B, Zhu WY, Zhao LX, Zhao GZ, Duan XW, Qin S, Li J et al. Serdani M et al. (2006). A five-gene phylogeny of (2012). Plantactinospora endophytica sp. nov., an Pezizomycotina. Mycologia 98: 1018–1028. actinomycete isolated from Camptotheca acuminata Steinberger Y, Whitford WG. (1983). The contribution of Decne., reclassification of Actinaurispora siamensis as rodents to decomposition processes in a desert Plantactinospora siamensis comb. nov. and emended ecosystem. J Arid Environ 6: 177–182. descriptions of the genus Plantactinospora and Sun MY, Dafforn KA, Johnston EL, Brown MV. (2013). Plantactinospora mayteni. Int J Syst Evol Micro 62: Core sediment bacteria drive community response to 2435–2442.

Supplementary Information accompanies this paper on The ISME Journal website (http://www.nature.com/ismej)

The ISME Journal